On Feb 18, 2007, at 10:10 PM, Daniel Leidert wrote:
CCing Blue-Obelisk list
...
as own "module" of blue-obelisk)? Such a central repository would also
allow a strict structuring of the data (valid/invalid formats, testing
of files with a wrong extension, ...). The test-implementations itself
(C/C++/XML/XSLT/Java/... code), should then be part of the related
projects (should be simple, to write the necessary code for ant/
make to
download the SVN tree of test files).
Yes, I think this is perfect. One minor note. Open Babel also accepts
some "non-standard" formats, so I'd suggest a structure like:
* valid standard (e.g., PDB 2.2 standard)
* non-standard (e.g., "PDB" files)
* invalid
I actually think it's good to make sure we have at least some "valid
standard" files and some invalid ones. The latter are really good for
testing code to make sure they're rejected. (Well, I also have my
nasty random binary dumps, which can wreck havok on C/C++ code. >:-)
I can probably add another large pile of files in addition to the
Open Babel set, so the combination should yield a pretty
comprehensive repository, IMHO.
Other thoughts?
-Geoff
_______________________________________________
Blue-obelisk mailing list
Blue-obelisk@hardly.cubic.uni-koeln.de
http://hardly.cubic.uni-koeln.de/mailman/listinfo/blue-obelisk